412 BOTANICAL GAZETTE [DECEMBER 
a toxic action seems hardly probable. It has been long known 
to physiologists that glycerin (5) penetrates the cell wall and 
the protoplast very quickly (6) and can be demonstrated in the 
cell sap. Thus plasmolysis is rapidly diminished. This speaks 
for the rapid penetration of the cell by glycerin and the question 
suggests itself whether the consequent loss of time to accommo- 
date itself to the decided osmotic changes brought about by the 
solution may not account in some measure for the greater harm- 
fulness seen in glycerin. This question, however, will receive 
further attention. 
In the case of sodium chlorid, one finds a different condition 
of things. As appears in the table of results, common salt, 
NaCl, plasmolyzes Spirogyra cells first at a concentration of % 
gm. mol., and first can be survived when containing ;1, gm. mol. 
only. If we assume for sodium chlorid a purely osmotic action, 
we are able, by making use of the action of cane sugar asa 
standard of osmotic effect, to calculate the corresponding con- 
centration of sodium chlorid that should just permit the survival 
of the alga. 
Since the molecules split up in large proportion into ions, 
each of which exerts an osmotic effect equal to that of a mole- 
cule not so split, it is necessary to have recourse to the expeti- 
mental results of physical chemistry. On the basis of the 
molecular conductivity of salt solutions, it is possible to calcu- 
late the percentage of salt molecules dissociated in a solution of 
a given strength. The electrical conductivity at the given vol- 
ume (#,) divided by the electrical conductivity at an infinite 
dilution (#,,) gives the degree of dissociation. Since the elec- 
trical conductivities are not worked out for all boundary concen- 
trations reached in this study, it has been found necessary to use 
the closest possible approximation. The error thus occasioned, 
however, is believed to be not sufficiently large to detract materl- 
ally from the usefulness of the results, since it is impossible to 
eliminate many minor inaccuracies when working on a question 
of this nature. 
By using tables of molecular conductivities of NaCl, brought 
